Infect Dis Obstet Gynecol 2001;9:221–225 Differentiation between women with vulvovaginal symptoms who are positive or negative for

Objective: To investigate whether clinical criteria could differentiate between women with vulvovaginitis who were culture positive or negative for vaginal Candida species. Methods: Vulvovaginal specimens were obtained from 501 women with a vaginal discharge and/or pruritis. Clinical information and wet mount microscopy findings were obtained. All specimens were sent to a central laboratory for species identification. Results: A positive culture for Candida species was obtained from 364 (72.7%) of the specimens. C. albicans was identified in 86.4% of the positive cultures, followed by C. glabrata in 4.5%, C. parapsilosis in 3.9%, C. tropicalis in 2.7% and other Candida species in 1.4%.Women with a positive Candida culture had an increased utilization of oral contraceptives (26.1% vs. 16.8%, p = 0.02) and antibiotics (8.2% vs. 0.7%, p = 0.001), and were more likely to be pregnant (9.1% vs. 3.6%, p = 0.04) than the culture-negative women. Dyspareunia was more frequent in women without Candida (38.0% vs. 28.3%, p = 0.03) while vaginal erythema (p = 0.01) was more common in women with a positive Candida culture. Conclusions: Although quantitative differences were observed, the presence of vaginal Candida vulvovaginitis cannot be definitively identified by clinical criteria.

It is difficult to obtain accurate information regarding the prevalence and incidence of vulvovaginitis associated with a Candida species infection. Although up to 75% of women will acknowledge having had a vaginal Candida infection during their lifetime 1 , this diagnosis is suspect. The skyrocketing sales of over-the-counter medications for Candida vaginitis, at a rate many times that of the number of infected women, highlights the prevalent overdiagnosis of this disorder. Many women, and unfortunately also many clinicians, label any vaginal discharge, itching, pain or burning as a 'yeast' infection. In three studies, more than half of the women with a supposed vaginal yeast infection were misdiagnosed [2][3][4] .
Conversely, in many women a true vaginal Candida infection may remain unrecognized. Detection of a vaginal Candida infection by microscopic examination of a vaginal specimen diluted in potassium hydroxide is relatively insensitive, especially for non-albicans species 3,5 . False-positive microscopic examinations are also possible 5 and are probably more common than generally suspected. It is also possible to have Candida vulvovaginitis with a false-negative Candida culture. At least 3000 organisms/ml are necessary to obtain a positive culture 6 .
In an attempt to more accurately characterize symptomatic women with a positive Candida species vaginal culture, and to differentiate them from women with vaginal symptoms due to other causes, a study was initiated in three cities in Brazil.

MATERIALS AND METHODS
This study was approved by the Clinical and Ethical Committee of Hospital das Clinicas, University of Sao Paulo, and informed written consent was obtained from all subjects. The study population consisted of 501 consecutive reproductive age women complaining of a vaginal discharge and/or vulvovaginal pruritis, seen as private patients in the Brazilian cities of Sao Paulo, Rio Grande de Sul and Salvador. Exclusion criteria included the use of immunosuppressive medications, vaginal medications or oral antifungal agents within the last 30 days.
Clinical and demographic data were collected at each center by a single participating physician. Signs and symptoms upon physical examination were standardized as much as possible between the different sites by providing common diagnostic criteria for erythema, edema, discharge and dysuria. Definitions were similar to those utilized by Eckert and colleagues 7 .
Specimens were obtained by scraping the vaginal walls with a cotton swab and immediately transferring the contents to a glass slide. A drop of saline was added and diagnosis of Candida was based on the observed presence of mycelium (branched hyphal elements) or blastospores (the unicellular yeast form). A second specimen was placed in transport medium and shipped to a central clinical laboratory. Specimens were cultured on Sabouraud agar containing chloramphenicol. Candida species were identified by the automated Amphotericin B (ATB) express method.
Comparisons between women with positive or negative Candida cultures for quantitative variables were analyzed by the Student's t-test for independent samples. The c 2 test was used to compare qualitative variables between both groups. Findings were considered significant at p < 0.05.

RESULTS
Candida was detected by culture in 364 (72.7%) of the subjects. The distribution of individual Candida species can be seen in Table 1. C. albicans was identified in 86.4% of the positive cultures, followed by C. glabrata (4.5%), C. parapsilosis (3.9%) and C. tropicalis (2.7%). A presumed identification of Candida was made by microscopic examination in 87.1% of women with a positive culture and in 5.1% of those with a negative culture. Trichomonas vaginalis was present in 2.9% and 1.4% of women with a negative and positive Candida culture, respectively. Clue cells were observed in 16.8% of women with a negative culture and in 9.1% of women with a positive Candida culture (p = 0.01).
For all analyses, the patients were divided into two groups based on the presence or absence of a positive Candida culture. Demographics of women in both groups are shown in Table 2. A higher percentage of black women, but not of white women or those of other races, were present in the culture-negative group (16.8%) than in the Candida culture-positive group (8.3%) (p = 0.01).
The relationship between predisposing factors and a positive or negative Candida culture is detailed in Table 3. Pregnancy (9.1% vs. 3.6%, p = 0.04), oral contraceptive usage (26.1% vs. 16.8%, p = 0.02) and current antibiotic use (8.2% vs. 0.7%, p = 0.001) were each associated with detection of a positive Candida culture. Conversely, a positive HIV serology (7.3% vs. 3.3%,  Table 1 Candida species identified by culture p = 0.05) and prior history of a sexually transmitted disease (17.5% vs. 9.7%, p = 0.01) were associated with a negative Candida culture. The HIV-seropositive women were at the earliest stages of their disease.
The relationship between patient-reported signs and symptoms and Candida culture findings is shown in Table 4. There was considerable overlap between the two groups in complaints of a vaginal discharge, vulvar pruritis and burning, dysuria and dyspareunia. Only dyspareunia was significantly different between the two groups: 38.0% in women with a negative culture vs. 28.3% in those with a positive Candida culture (p = 0.03).
Clinical findings in the patient groups are shown in Table 5. Although there was an overall high degree of similarity between subjects regardless of their Candida culture status, women with a positive culture had a higher prevalence of vaginal erythema (p = 0.01).

DISCUSSION
Although some quantitative differences in the frequency of patient symptoms, clinical findings and predisposing factors were identified between groups of women who were culture-positive or culture-negative for Candida species, none of the evaluated criteria were pathognomonic for Candida in the vagina. Similar findings have been reported previously by others 2,6-8 . In a large study of women attending a sexually transmitted disease clinic, only 28% of 545 women with pruritis, burning or a vaginal discharge were C. albicansculture positive 7 . Thus, in symptomatic women, a positive wet mount or culture for Candida is necessary to assess whether this organism is present in the vagina.
The findings in the present study of associations between a positive Candida culture in symptomatic women and current oral contraceptive and antibiotic usage and pregnancy parallel earlier reports [9][10][11] . We recognize that other studies have     reported that women using oral contraceptives containing estrogen levels of 35 mg or less did not have an increased rate of candida vulvovaginitis 7,11,12 . This does not seem to be true, however, for our study population. The oral contraceptives used by the patients examined contained estrogen levels below 35 mg.
The etiology of the vaginal symptoms in the majority of our 137 Candida culture-negative patients remains undetermined. Bacterial vaginosis was a possible cause in 23 of these women, based on detection of clue cells by microscopy. An additional 4 women were positive for T. vaginalis by microscopic examination. Other possible causes of the observed symptoms, not evaluated in the present study, include allergic vaginitis, papillomavirus infection, desquamative vaginitis, cervicitis or estrogen deficiency. In addition, we acknowledge that in the Candida positive group, clinical signs and symptoms in at least some of the women may have been due to causes other than the presence of Candida species. However, the strong association between culture and wet mount suggests the presence of a high Candida concentration in the majority of the positive women and, therefore, an increased likelihood of Candidarelated symptomatology.
Host factors must also be taken into consideration when attempting to diagnose vulvovaginitis. It has been reported that there is no association between the Candida concentration in the vagina and clinical symptoms 13 . Women with Candida counts as low as 100 organisms/ml may be highly symptomatic while some women with vaginal Candida concentrations as high as 10 000/ml may be asymptomatic. Thus, women who are allergic to Candida antigens or products 14 can become symptomatic even when vaginal concentrations of this organism are below the level detectable by culture. Furthermore, a low level of Candida in the vagina can synergize with histamine released in response to other allergens to induce prostaglandin E2 production 15 . This, in turn, inhibits the cellmediated immune response necessary to prevent Candida proliferation.
It is apparent that patient symptomatology, clinical examination and medical history are insufficient to distinguish vulvovaginitis associated with the presence of a Candida species from vulvovaginitis due to other causes. Initiation of antifungal treatment based solely on these criteria will be ineffective for the many women who are negative for this microorganism. In symptomatic women, a positive wet mount or culture for Candida species is necessary to determine whether this microbe is present.